Myoelectric Prosthetic Arm Product

Overview

Myoelectric prosthetics are motor-driven upper-limb prostheses controlled via surface electromyography (EMG), measuring subtle electrical signals from muscles in the residual limb. When a user contracts their biceps or triceps muscles, the EMG Electrode Array detects the voltage change (typically 10–500 microvolts), which is amplified by the EMG Instrumentation Amplifier, filtered, digitized by the Analog-to-Digital Converter, and decoded by the Main Processor MCU into a specific motor command—open hand, close hand, rotate wrist. This intuitive control paradigm leverages the user's remaining motor intent, enabling voluntary, proportional grasp force and multi-articulate finger control that passive mechanical prostheses cannot match.

The defining advantage is voluntary motor control without mechanical links: the user thinks "grip," not "flex wrist," and the prosthesis executes a learned motor pattern. This reduces cognitive load and allows simultaneous hand and wrist movement, approximating natural bimanual coordination. Modern myoelectric arms also provide haptic feedback—the Grip Force Transducer measures grip force and can drive vibration alerts when approaching target force, closing the sensorimotor loop.

EMG Acquisition and Signal Processing

The myoelectric-prosthetic-arm-electrode-array uses dual silver-chloride surface electrodes placed on skin over the biceps and triceps, with a common reference electrode on the wrist. As muscle fibers depolarize during contraction, the potential difference between electrodes oscillates at frequencies dominated by 20–250 Hz motor unit action potentials. The raw EMG signal is picked up at 10–100 microvolts and carries substantial noise: 60 Hz mains interference, motion artifacts, and cross-talk from nearby muscles.

The Signal Processor and Motor Driver on-board PCB performs analog conditioning in three stages:

1. Instrumentation amplifier (INA128 or equiv.): Amplifies the differential EMG signal 1000–10000 times while rejecting 60 Hz common-mode interference by >80 dB using high input impedance (>10 GΩ) and balanced input resistors.

2. Analog filtering: A Analog Filter Network removes DC baseline drift with a high-pass corner at 100 Hz and attenuates muscle-contraction artifacts and EMG noise above 500 Hz with a low-pass filter.

3. Analog-to-digital conversion: A Analog-to-Digital Converter samples the filtered signal at 2 kHz per channel, buffering 1–2 seconds of history into the Main Processor MCU main memory.

Real-Time Pattern Recognition

The EMG Feature Extraction computes time and frequency-domain features every 50 milliseconds: root-mean-square (RMS) amplitude, zero-crossing rate, and power spectral density moments. These features capture the "shape" of each muscle contraction. Different muscle groups (biceps, triceps, brachialis) and contraction intensities produce distinct feature vectors.

A Pattern Classifier (typically linear discriminant analysis or a shallow neural network with 1–2 hidden layers and 16–32 neurons) maps feature vectors to predicted grip commands: open hand, power grip, pinch grip, lateral (key) grip, and wrist rotate-clockwise or counterclockwise. The classifier is trained during user fitting—the prosthetist cues the user to perform each movement 5–10 times, collecting labeled feature vectors. At runtime, the classifier outputs a probability distribution over grip classes; the Main Processor MCU selects the command with highest confidence if it exceeds a tunable threshold (typically 70–80%).

Motor Control and Grasp Execution

When the classifier selects "close-hand," the Dual Motor Driver IC energizes the Hand Drive Motor (a small DC motor with 50:1 planetary reduction), which winds a Flexor Tendon Cable connected to the finger linkages. As the cable pulls, the Finger Linkage Assembly flexes all four fingers and opposable thumb through a four-bar mechanism or pulley system, forming a power grip. The speed of motor rotation is proportional to EMG signal RMS: a weak contraction may open fingers slowly (allowing dexterous task release), while a strong contraction opens them quickly.

Grip force is monitored by the Grip Force Transducer, a 0–50 N load cell embedded in the palm. When grip force exceeds the user's tuned threshold (often set to 40% of max to prevent user fatigue), the motor saturates and holds load without further rotation. Some advanced systems integrate haptic feedback: a vibration motor in the socket alerts the user when grip force reaches 80% of max, preventing overgriping and crushing fragile objects.

Wrist rotation is independent: the Wrist Servo Motor and Wrist Reduction Gearbox provide 100:1 mechanical advantage, enabling 180° pronation/supination (palm-up to palm-down) with high torque. The Wrist Angle Sensor (potentiometer or hall-effect encoder) feeds back wrist angle; the processor locks the wrist at the user's selected angle, preventing unwanted rotation during grasp.

Power Management and Battery

The Power Battery and Charger consists of two LiPo cells in parallel (dual 3.7 V, 2 Ah each → 7.4 V, 4 Ah nominal), providing ~30 Wh energy. Hand motor operation draws 300–800 mA depending on load and duty cycle; wrist servo draws 100–200 mA during rotation and nearly zero when locked. A typical day (8 hours of continuous walking and light object manipulation, ~100 grasp-release cycles) consumes 20–25 Wh, leaving one day of reserve before recharge. The Charge Management IC manages charging via standard USB, balancing the two parallel cells and limiting inrush current to 1 A.

Socket Fitting and User Adaptation

The custom Socket and Interface Module is critical to EMG signal quality. The Custom Socket Liner is thermoformed or hand-molded from EVA foam or medical-grade silicone to precisely fit the user's residual limb contours. The myoelectric-prosthetic-arm-electrode-array is embedded in this liner at positions determined by palpating biceps and triceps motor points, ensuring good electrical contact with skin.

During the first week of use, the User Adaptation Module continuously monitors user performance: each time the user attempts a command and the prosthesis executes the correct action, the classifier is rewarded (confidence threshold adjusted downward); each time it misclassifies, the threshold is adjusted upward. This online learning substantially improves accuracy within 3–7 days of daily use, reducing false opens/closes and fatigue.

Cosmetic Design and Glove

The hand and wrist are encased in a cosmetic Hand Skeletal Frame glove of silicone, thermoplastic polyurethane, or fabric, tinted to match the user's skin tone. The glove covers the motor, gearbox, and electronics, presenting a humanoid appearance. Some advanced models embed conductive gel pads under the glove for capacitive touch sensing, enabling the prosthesis to detect when fingers contact objects and automatically modulate grip force.

Clinical Outcomes and Limitations

EMG-controlled prosthetics substantially improve functional outcomes compared to body-powered prostheses: users report higher task success rates (75–95% vs. 40–60% on precision tasks), faster completion times, and higher satisfaction due to reduced fatigue. However, EMG control is not perfect: signal quality degrades with sweat, electrode shift during long-term wear, and muscle fatigue. Some users require 1–2 months of training before achieving comfortable control. The technology is most successful in transhumeral (above-elbow) amputees with preserved muscle bulk; transcarpal (wrist disarticulation) and partial-hand users may find myoelectric prosthetics less practical due to fewer muscle sites and space constraints.